Risk-relevant plants like Oil & Gas (O&G) or nuclear ones are subjected to strict safety regulations. Risk Assessment is mandatory for these plants, and the damage quantification is a crucial step which has to be carefully addressed.

Nowadays the state of practice for consequences estimation entails the use of semi-empirical methods which permit a fast evaluation of the large number of accidental scenarios needed for a Quantitative Risk Assessment (QRA).

However, in case of large and congested industrial environments like offshore platforms or equipment inside nuclear primary containment buildings, the aforementioned methods usually lead to an overestimation of the damage areas, for example because they neglect the space congestion which highly affects the accident evolution. A more accurate analysis can be performed using the Computational Fluid Dynamics (CFD), nonetheless its high computational cost represents an important drawback.

In this work a CFD approach, called Source Box Accident Model (SBAM), is presented. It models high-pressure gas releases (>10 bar) in congested environments guaranteeing a good computational cost-accuracy compromise. The aim of SBAM is to permit a fast consequences estimation (evaluation of flammable/toxic areas, etc.) via CFD, in order to have a simulations time compatible with the plants design schedule. The long-term objective is to integrate the CFD contribution in a safety driven design process.

SBAM is based on the splitting of the multi-physics and multiscale phenomena characterizing the accident: the initial supersonic compressible release and the successive low speed dispersion. The first one is simulated in a small domain called Source Box (SB) and the second one in the case study domain.

The two simulations are coupled in a suitable way, through proper parameters which are extensively discussed. This work presents a detailed description of SBAM and two different analyses: a sensitivity study on the coupling parameters and a numerical benchmark which uses a standard CFD simulation as reference. The sensitivity analysis shows that the coupling is a crucial step of the method and the coupling parameters must be treated in the most accurate way. The numerical benchmark shows that SBAM is not introducing significant errors with respect to a standard CFD simulation and, in addition, permits a relevant simplification in the simulation setup and computational cost reduction.

与风险相关的工厂，例如石油和天然气（O＆G）或核电厂，都受到严格的安全法规的约束。这些工厂必须进行风险评估，而损害量化是至关重要的一步，必须谨慎处理。

如今，后果评估的实践状态需要使用半经验方法，该方法可以快速评估定量风险评估（QRA）所需的大量意外情况。

但是，在大型且拥挤的工业环境中，例如海上平台或核一级安全壳内部的设备，上述方法通常会导致对损坏区域的高估，例如，因为它们忽略了严重影响事故演变的空间拥堵。使用计算流体动力学（CFD）可以执行更准确的分析，尽管如此，其高昂的计算成本还是一个重要的缺点。

在这项工作中，提出了一种称为源箱事故模型（SBAM）的CFD方法。它在拥挤的环境中对高压气体释放（> 10 bar）进行建模，从而确保了良好的计算成本准确性折衷方案。SBAM的目的是通过CFD进行快速的后果评估（易燃/有毒区域的评估等），以使仿真时间与工厂设计时间表兼容。长期目标是将CFD贡献整合到安全驱动的设计过程中。

SBAM基于表征事故的多物理场和多尺度现象的分裂：初始的超音速可压缩释放和连续的低速弥散。第一个在称为Source Box（SB）的小域中进行仿真，第二个在案例研究域中进行仿真。

通过广泛讨论的适当参数，两种模拟以合适的方式耦合。这项工作提供了对SBAM的详细描述和两个不同的分析：对耦合参数的敏感性研究和使用标准CFD模拟作为参考的数值基准。灵敏度分析表明，耦合是该方法的关键步骤，必须以最准确的方式处理耦合参数。数值基准表明，SBAM不会在标准CFD仿真方面引入重大错误，此外，它还可以简化仿真设置并降低计算成本。